From Gene to Function in Dystonia

DMRF/NINDS Dystonia Workshop - "From Gene to Function in Dystonia"
Chairs: Xandra Breakefield, Laurie Ozelius, Phyllis Hanson, William Dauer
January 19 - 21, 2001
Phoenix, AZ

Overview: Xandra Breakefield
The discovery in 1997, of the TORA1 (DYT1) gene, responsible for early onset torsion dystonia, has catapulted dystonia from the realm of neurology and human genetics into protein structure/biochemistry, cell biology and animal models. Unusual aspects of the disease include: the childhood window of susceptibility during a period of motor learning; low penetrance of a dominant disease gene; and apparent lack of frank neurodegeneration. These suggest a novel mode for neuronal dysfunction with important implications in basic neuroscience. A large number of genes (up to 14) underlie various forms of the hereditary dystonias and indicate a broad and dynamic field of players subserving the control of fine movement within the basal ganglia. Biochemists are intrigued as the encoded protein, torsinA, is one of the few members of the AAA+ chaperone family of proteins involved in a human disease and apparently the first to reside in the endoplasmic reticulum (ER). In fact, the torsin gene family first appears in evolution with the emergence of multicellular organisms. The mutation (GAG deletion) responsible for most cases of early onset torsion dystonia is predicted to disrupt ring formation necessary for protein function. The tendency for the mutant protein to form large, whorled, ball-like inclusions when expressed at high levels in cells and the apparent involvement of one of the nematode homologues in nuclear rotation suggest a role for the torsins in membrane dynamics critical to neuronal development and function.

Several themes were raised throughout the meeting, which may provide clues for torsin function. First, many convergent lines point to a relationship between dystonia and Parkinson's disease (PD). This includes the fact that many forms of dystonia present with parkinsonian features and, conversely, that early onset PD frequently presents initially as dystonia. Lewy bodies found in the brains of PD patients contain both a-synuclein and torsin. Dopaminergic neurons in the substantia nigra have a critical role in both diseases and express high levels of torsin A and a-synuclein, which has been implicated in the etiology of PD. Homozygous mice with the A53T alpha synuclein mutation manifest with dystonic-like movements. A second theme was that defects in mitochondrial function due to mutations in both mitochondrial and nuclear genes can cause dystonia, as attested by Leber's optic atrophy and deafness-dystonia. Interestingly, the mammalian protein with highest homology to the torsins, SKD3, is a mitochondrial protein that increases in response to oxidative stress. Third, signs point to a number of neuronal types which may contribute to the manifestation of dystonia, including expression of torsin in select neurons in the basal ganglia, substantia nigra, cerebellum and hippocampus. Levels of torsin are upregulated in striatal interneurons during a window of development associated with high synaptogenesis, and torsin is found in the adult primate brain at symmetric synapses in the basal ganglia. Understanding how torsin functions in neuronal dynamics will be critical to resolving the disease etiology and possibly in understanding aspects of motor development and function. Torsin may affect membrane trafficking involved in neuronal polarity and integrity needed for synaptic communication.

Genetics - Laurie Ozelius
It was clear from this meeting that the discovery of the gene for early onset dystonia has led to an explosion of work on the TOR1A protein and its role in disease and underscores the need to find other genes responsible for the various forms of dystonia.

With regard to TOR1A, (Laurie Ozelius, Albert Einstein University) a new 18 bp deletion mutation was described that removes 6 amino acids from the C-terminus of the protein including a putative phosphorylation site. In addition, studies aimed at determining genetic factors that might influence the penetrance of early onset dystonia were discussed. Using the simplest model of inheritance (autosomal dominant with a high gene frequency), six chromosomal regions of positivity (multiple positive markers) have been identified on chromosomes 2, 5, 7 (two regions) and 8 (two regions). One of the regions on chromosome 7 and another on chromosome 8 overlap with regions that have been implicated in myoclonus dystonia and adolescent onset dystonia (DYT6), respectively. If genes influencing penetrance can be identified, this may lead to new treatments aimed at delaying or preventing the manifestation of the mutant TOR1A gene in carriers.

Two forms of dystonia, Leber's optic atrophy (Doug Wallace, Emory) and deafness-dystonia (Craig Blackstone, MGH) are caused by mutations in the mitochondrial and nuclear genes, respectively, which encode proteins critical to mitochondrial function. In Leber's, a heterozygous state of a particular mutation in the mitochondrial genome causes blindness, while a homozygous state leads to childhood-onset, generalized dystonia,with varying phenotypes mediated by degrees of blockage in oxidative phosphorylation. The deafness-dystonia gene (nuclear) encodes a protein needed for generation and division of mitochondria within the cell. Overexpression of a mutated form of an interacting gene leads to whorled, membrane inclusions, as seen with mutant torsinA, and long tubular mitochondria.

Several other forms of dystonia are associated with parkinsonism, dopa-responsive dystonia (DRD) is phenotypically similar to early onset dystonia; however some individuals become parkinsonian later in life. This form of dystonia shows a dramatic therapeutic response to small doses of L-dopa which resolves both the dystonic and parkinsonian, symptoms without the side effects usually associated with PD. Heterozygous mutations in GTP cyclohydrolase 1, the rate limiting enzyme in the formation of the co-factor tetrahydrobiopterin, or homozygous mutations in tyrosine hydroxylase, the rate limiting enzyme in the production of dopamine from tyrosine, both cause DRD. Treatment with L-dopa bypasses both of these blocks to the pathway, thus allowing the patients to make adequate levels of dopamine. Rapid onset dystonia parkinsonism is a rare form of dystonia characterized by acute onset usually in hours to days of either dystonia, parkinsonism or both syndromes. Patients with this disorder have only very slight benefit from L-dopa, although levels of the dopamine metabolite, homovanillic acid, is decreased in at least some of the patients.

The gene for this disorder has been linked to chromosome 19 but has yet to be identified. Finally, the X-linked form of dystonia known as Lubag (Andrew Singleton, Mayo Jacksonville) is found only in the Capiz province of the Philippines due to a founder mutation. This form of dystonia usually begins focally but rapidly progresses to generalized dystonia with about half the cases also having parkinsonism. This is the only form of hereditary dystonia known to have degeneration in the brain, including regions of the basalganglia. The identification of the gene for this form of dystonia is being actively pursued.

Other global issues that were discussed mainly focused on the collection of patient material. These included a collaborative effort among investigators to collect and share families with later onset dystonia, to develop a registry of dystonia patients and families that can be shared among researchers, and to coordinate efforts to involve patients and family members in brain donations. Particularly in light of the penetrance issues, the brains of non-manifesting carriers of TOR1A may provide clues as to specific morphologic features related to the disease state.

Torsins and chaperone proteins -Phyllis Hanson
Cell biologists and biochemists are working toward understanding what the protein encoded by the TOR1A gene normally does in the cell and how a single amino acid deletion can change its activity and thereby cause early-onset torsion dystonia. The session on torsins and chaperone proteins at the recent DMRF workshop highlighted progress in this area, and generated enthusiastic discussions making it clear that further developments will not be long in coming.

TorsinA belongs to the emerging and essential family of AAA+ ATPases. These enzymes appear to function in a chaperone-like manner, carrying out noncovalent conformational modification of stable proteins and protein-protein complexes. TorsinA is the first AAA+ ATPase localized to the lumen of the secretory pathway. Based on studies of other AAA+ ATPases, including the membrane fusion ATPase, NSF, and the bacterial protein folding chaperone, ClpB, one can hypothesize that torsinA is a chaperone with specific target(s) in the secretory pathway. Other AAA+ ATPases function as hexameric ring-shaped oligomers, and are only active in their oligomeric state. If torsinA shares these properties, this may explain how a mutation in it would generate a dominantly transmitted phenotype.

This session brought together investigators with diverse backgrounds and interests, addressing questions that ranged from how to produce and study torsinA using biochemical and high resolution structural approaches, to where and how torsinA may function normally in cells and in the brain as a whole. In all cases, efforts to understand how the GAG deletion alters torsinA structure and function were emphasized.

Specific progress included descriptions by Phyllis Hanson (Washington Univ.) and Gordon Rule (Carnegie Mellon) of recombinant protein production for biochemical and structural analysis. Initial attempts in E. coli were only partially successful, and both have now turned to developing eukaryotic expression systems with which to generate active protein. Michal Zolkiewski (Kansas State) joined Hanson in describing AAA+ enzymes that can serve as biochemical and structural paradigms for torsinA, and highlighted the critical role likely to be played by the C-terminal domain of torsinA that contains the GAG deletion.

Benjamin Cravatt (Scripps) characterized the localization and cellular effects of torsinA overexpressed in mammalian cells, showing that torsinA is indeed a glycoprotein co-localized predominantly with markers of the endoplasmic reticulum. Interestingly, GAG deleted torsinA exhibits a strikingly different localization to large spheroid structures. Xandra Breakefield (MGH) used electron microscopy to show that these structures consist of membrane whorls, probably derived from the endoplasmic reticulum. Breakefield showed studies of endogenous torsinA in neuronal cell lines, again demonstrating co-localization with markers of the ER as well as the secretory pathway. She also described the minimal effects of various cell stress paradigms on torsinA expression. P. Shashidaran (Mt. Sinai) described a variety of studies of torsinA cell biology, including a clear demonstration by isoelectric focussing that DYT1 mutant patients express both wild-type and mutant protein in neuronal tissue. Craig Blackstone (MGH) completed the cell biology presentations by describing studies of a protein underlying a non-DYT1 form of dystonia, dystonia-deafness, and described its interactions with the mitochondrial dynamin-like protein, DVLP. Interaction of these two proteins is required for proper mitochondrial fissioning. These studies may provide insight into new ways in which mitochondrial dysfunction can give rise to dystonia.

One of the big mysteries in understanding how a mutation in a protein such as torsinA generates a focal neurological disease is why particular neurons are more susceptible to effects of the mutation than others. A possible explanation for such specificity may be found in the expression of this protein at different levels in specific neurons and at particular times during development. Marie-Francoise Chesselet (UCLA) described immunolocalization studies looking at torsin protein expression over the course of development in the rat, noting a striking increase in the expression of torsinA in striatal interneurons at around P14, the time of the highest level of synaptogenesis in the basal ganglia. This raises the interesting possibility that some of torsinA's functions might be critical during neuronal development. David Standaert (MGH) described in situ hybridization studies, as well as antibody localization studies of torsinA in human control brains. Using a monoclonal antibody he noted the interesting finding that although torsinA expression in the adult human brain is highest in dopaminergic neurons, very little immunoreactive protein can be found in their cell bodies, with most appearing in vesicular structures in axons and synaptic terminals. Finally, Nutan Sharma (MGH) described FRET techniques for improving the resolution of double labelling studies in the light microscope, holding out promise for refinement in studies of co-localization of torsinA with potential interacting partners including a-synuclein, also noted by Dr. Shashidaran.

Animal models - William Dauer
The animal models section of the meeting focused on three general themes: 1) general issues of animal model construction and analysis, 2) etiologic specific models of DYT1 primary dystonia and 3) animal models of secondary dystonia.

Dr. Don Price (Johns Hopkins), Dr. Michael Lee (Johns Hopkins), Dr. Yuquig Li (University of Illinois), and Dr. Alan Peterson (Royal Victoria Hospital) discussed different examples of animal model construction and analysis. Don Price presented on animal models of Alzheimer's disease and highlighted how etiologic specific mouse models can lead to a fundamental molecular understanding of disease pathogenesis, and suggest rational therapeutic targets. He discussed how these mice have helped to elucidate the interactions between presenilin 1 (PS1) and amyloid precursor protein (APP), and the crucial role of beta secretase in APP processing. Importantly, beta secretase knock out mice do not produce any of the normal APP products, a result which suggests that inhibitors of this protein should slow or prevent the production of amyloid plaques and Alzheimer's disease. Michael Lee presented his work on alpha synuclein transgenic mice, which supported a specific pathogenic effect of the A53T form of mutant alpha synuclein. Mice that overexpress the A53T mutant alpha synuclein develop a progressive motor phenotype consisting of decreased movement, poor balance and dystonia which culminates in early death, but mice that overexpress comparable levels of wild type or A30P mutant alpha synuclein do not develop these abnormalities. The A53T mutant mice develop a progressive astroglial reaction of brainstem and cerebellar nuclei. Yuquig Li and Alan Peterson discussed the novel approaches they are using to generate torsinA mutant models. Yuquig Li detailed his plans to use various lines of Cre recombinase expressing mice to generate region specific knock in mice that express mutant torsinA specifically in the striatum, as well as plans to generate tissue specific knock out mice. Alan Peterson presented a method which allows one to insert a transgene at a defined chromatin site highly permissive for expression of transgenes. This is accomplished by cloning the desired transgene in a genomic vector containing a portion of the HPRT gene, and using homologous recombination to target this vector in ES cells that lack the vector HPRT sequence in one allele. Homologous recombination thus restores HPRT function to the clone, placing the transgene at a defined site. This location has been shown to allow promoters to express in a physiological pattern, and the technique allows precise structure/function studies to be performed that are difficult or impossible using classical transgenic techniques. Alan Peterson plans to make a number of transgenic torsinA lines with this technique, including ones that express torsinA widely, or only in developing neurons.

Etiologic specific models of DYT1 primary dystonia were discussed by Scott Emmons (Albert Einstein), Lesilee Rose(UC at Davis), Jim Gusella (MGH), and William Dauer (Columbia CPS). Emmons and Rose discussed studies in C. elegans. Emmons focused on two torsin related proteins (trp), trp1 and trp2, and another torsin related protein, OOC-5. Using a GFP-tagged transgene, he found trp1 to be expressed in a subset of neurons, muscles and gut cells, and work to define the expression pattern of trp2 is underway. Another set of experiments explored the consequences of blocking trp gene function (using the RNAi technique), and showed that blockade of OOC-5 resulted in an embryonic arrest phenotype, while blockade of trp1 gave no phenotype. Work by Dr. Rose has shown that the torsin-related OOC-5 gene and another gene, OOC-3, are required for normal polarity of a particular cell during the asymmetric divisions of the early C. elegans embryo. Disruption of these genes leads to a specific polarity defect involving an altered distribution of the PAR-3 and PAR-2 proteins, which are required for normal polarity of the early C. elegans embryo. Additionally, her lab has found OOC-5 protein to be localized to all cells of the early embryo, in a pattern reminiscent of the ER and/or secretory pathway in other cell types. Others have reported that the OOC-3 protein is a predicted transmembrane protein localized to the ER. Rose and her colleagues find that OOC-5 and OOC-3 proteins colocalize extensively in early embryos. These localization studies are consistent with the proposed ER localization of human torsinA, suggesting that some essential ER-related function has been conserved throughout evolution in the torsin proteins. Furthermore, it is possible that OOC-3 encodes a protein that interacts with the OOC-5 gene product, and may thus represent a homolog of a torsinA-interacting protein.

Gusella presented his work on torsinA mutants in Drosophila. There is only a single torsin homolog in Drosophila, termed torsin related protein 4 (TORP 4). Null mutants of TORP 4 have been limited by the fact that there is a P element hot spot very close to the TORP 4 gene. In addition, the specific perturbation of TORP4 is likely to be quite challenging, as it is in close proximity to 4 other genes. Overexpression of TORP 4 in the Drosophila eye produces a "rough" phenotype, and abnormalities in the photoreceptor and pigment layer. Work in progress is aimed at expressing wild type and mutant human torsinA in Drosophila, and knocking out TORP 4 by using homologous recombination. It is anticipated that the eye phenotype may be used to screen for TORP 4 interacting or modifying proteins, and similar screening strategies will also be valuable in the C. elegans work described by Emmons and Rose.

In the final talk relating to DYT1 models William Dauer (Columbia) described the generation and characterization of torsinA mutant mice. "Knock down" mice expressing low levels of mutant torsinA (and no wild type protein) display deficient habituation in open field studies. Cyclic voltametry studies measuring striatal dopamine release in these mice show that the magnitude of release and the kinetics of dopamine uptake and vesicle recycling are normal in these mice, as are the levels of the D1 receptor and dopamine transporter. Mice that contain the pathogenic GAG deletion in their native torsinA gene (knock in mice) have not yet been found to have an abnormal behavioral phenotype, and studies of dopamine receptor and transporter level also appear normal in this line. TorsinA knock out mice die at P0, presumably because of a failure to feed, although the precise reason for this inability to feed has not been clarified.

Mark LeDoux completed the animal models section with a talk on the dystonic rat, a naturally occurring autosomal recessive mutation. These rats display dystonic symptoms by 12 days of age, and die from progressive motor disability by approximately 35 days. Biochemical studies have shown increased GAD activity, decreased muscimol binding, and increased glucose utilization within the cerebellar nuclei. These findings suggest increased activity at the level of Purkinje cell synapses within the cerebellar nuclei. In contrast studies of the basal ganglia including those of muscarinic cholinergic, dopamine D2, 5HT2, and benzodiazepine receptor binding have been normal. Behavioral studies with harmaline and electrophysiologic studies of Purkinje cell firing suggest that a defect in the pathway from the inferior olive to climbing fiber synapses on cerebellar Purkinje cells is the primary abnormality in the dystonic rat. These studies support a role for non-basal ganglia related structures in dystonic physiology.